In the course of completing our study, Stroke Prevention in Atrial Fibrillation: Impact of Mental Illness (IIR 04- 248), we learned that half of Veterans receiving warfarin in VA actually received INR testing in Medicare (MC) during Fiscal Year 2004 (FY04), suggestion that some anticoagulation care was occurring outside VA and that Veterans were engaging in DU of VA and Medicare anticoagulation services. Moreover, patients with INR testing in VA+MC had lower unadjusted frequencies of anticoagulation process measures, suggesting that dual care was not supplementary but may be causing harm. Failure to coordinate VA AF patients' care could increase risk for strokes, bleeds and death. Our study investigates whether patients treated with warfarin who receive care in both VA and Medicare have poorer outcomes. We also examine whether dual users are less likely to receive regular INR tests. If dual use is associated with adverse outcomes, important implications could follow: 1. Patients might consider using a single institution of all their AF care. 2. Clinicians might heighten their vigilance about monitoring anticoagulation care in dual users. 3. Health care systems might consider implementing systems that optimize coordination of services between VA and the private sector. We propose the following objectives in studying Veterans with AF receiving warfarin in VA: 1. To characterize patterns of Dual Use (VA+MC) 2. To determine the effect of DU on processes of anticoagulation care 3. To determine the effect of DU on outcomes of AF care (stroke, hemorrhage, death) We will apply approaches developed by our team for IIR 04-248 to create a cohort of Veterans VA users with prevalent AF on warfarin therapy during the observation period. A major strength of our plan is that we can apply methods and protocols that we developed for IIR 04-248 that are specialized to AF research using VA and CMS data (e.g., for merging databases and for identification of AF, outcomes of AF care); while assembly of these multiple complex databases using updated data sources will still be labor-intensive, our prior experience will improve our efficiency, making it possible for us to propose a markedly shortened (2 -year) project timeline. In this way we will leverage VA resources by building on prior work to answer critical questions about ramifications of today's heterogeneous U.S. health care delivery system. Data will be drawn from existing sources: VA's National Patient Care Database, VA's Decision Support System, VA's Vital Status file, and Medicare data. Varying definitions of DU will be examined, with a focus on DU based on receipt of INR monitoring in VA and Medicare. Strokes and bleeds will be identified from ICD9 codes using existing algorithms. Death will be identified from VA's Vital Status file. INR monitoring will be identified from lab and CPT codes, and Time in Therapeutic Range from INR values.